Commit | Line | Data |
---|---|---|
eaaeae39 NR |
1 | package be.nikiroo.gofetch.support; |
2 | ||
3 | import java.io.IOException; | |
4 | import java.io.InputStream; | |
5 | import java.net.URL; | |
6 | import java.util.ArrayList; | |
7 | import java.util.List; | |
8 | ||
9 | import org.jsoup.helper.DataUtil; | |
10 | import org.jsoup.nodes.Document; | |
11 | import org.jsoup.nodes.Element; | |
12 | import org.jsoup.select.Elements; | |
13 | ||
14 | import be.nikiroo.gofetch.data.Comment; | |
15 | import be.nikiroo.gofetch.data.Story; | |
16 | ||
17 | /** | |
18 | * Support <a href='https://lwn.net/'>https://lwn.net/</a>. | |
19 | * | |
20 | * @author niki | |
21 | */ | |
22 | public class LWN extends BasicSupport { | |
23 | @Override | |
24 | public String getDescription() { | |
25 | return "LWN: Linux Weekly Newsletter"; | |
26 | } | |
27 | ||
28 | @Override | |
29 | public List<Story> list() throws IOException { | |
30 | // TODO: comments + do not get comment for [$] stories | |
5c056aad | 31 | |
eaaeae39 NR |
32 | List<Story> list = new ArrayList<Story>(); |
33 | ||
34 | URL url = new URL("https://lwn.net/"); | |
35 | InputStream in = open(url); | |
36 | Document doc = DataUtil.load(in, "UTF-8", url.toString()); | |
37 | Elements stories = doc.getElementsByClass("pure-u-1"); | |
38 | for (Element story : stories) { | |
39 | Elements titles = story.getElementsByClass("Headline"); | |
40 | Elements listings = story.getElementsByClass("BlurbListing"); | |
41 | if (titles.size() == 0) { | |
42 | continue; | |
43 | } | |
44 | if (listings.size() == 0) { | |
45 | continue; | |
46 | } | |
5c056aad | 47 | |
eaaeae39 NR |
48 | Element listing = listings.get(0); |
49 | if (listing.children().size() < 2) { | |
50 | continue; | |
51 | } | |
eaaeae39 NR |
52 | |
53 | String title = titles.get(0).text(); | |
54 | String details = listing.children().get(0).text(); | |
25271075 NR |
55 | String body = ""; |
56 | // All but the first and two last children | |
5c056aad | 57 | for (int i = 1; i < listing.children().size() - 2; i++) { |
25271075 NR |
58 | Element e = listing.children().get(i); |
59 | body = body.trim() + " " + e.text().trim(); | |
60 | } | |
61 | body = body.trim(); | |
5c056aad | 62 | |
eaaeae39 NR |
63 | String author = ""; |
64 | int pos = details.indexOf(" by "); | |
65 | if (pos >= 0) { | |
66 | author = details.substring(pos + " by ".length()).trim(); | |
67 | } | |
5c056aad | 68 | |
eaaeae39 NR |
69 | String date = ""; |
70 | pos = details.indexOf(" Posted "); | |
71 | if (pos >= 0) { | |
72 | date = details.substring(pos + " Posted ".length()).trim(); | |
73 | } | |
eaaeae39 NR |
74 | |
75 | String id = ""; | |
76 | String intUrl = ""; | |
77 | String extUrl = ""; | |
78 | for (Element idElem : story.getElementsByTag("a")) { | |
79 | // Last link is the story link | |
80 | intUrl = idElem.absUrl("href"); | |
81 | pos = intUrl.indexOf("#Comments"); | |
82 | if (pos >= 0) { | |
5c056aad | 83 | intUrl = intUrl.substring(0, pos - 1); |
eaaeae39 NR |
84 | } |
85 | id = intUrl.replaceAll("[^0-9]", ""); | |
86 | } | |
87 | ||
5c056aad NR |
88 | list.add(new Story(getType(), id, title, details, intUrl, extUrl, |
89 | body)); | |
eaaeae39 NR |
90 | } |
91 | ||
92 | return list; | |
93 | } | |
94 | ||
95 | @Override | |
5c056aad | 96 | public void fetch(Story story) throws IOException { |
eaaeae39 | 97 | /* |
5c056aad NR |
98 | * URL url = new URL(story.getUrlInternal()); InputStream in = |
99 | * open(url); Document doc = DataUtil.load(in, "UTF-8", url.toString()); | |
100 | * Elements listing = doc.getElementsByTag("main"); if (listing.size() > | |
101 | * 0) { comments.addAll(getComments(listing.get(0))); } | |
102 | */ | |
eaaeae39 NR |
103 | } |
104 | ||
105 | private List<Comment> getComments(Element listing) { | |
106 | List<Comment> comments = new ArrayList<Comment>(); | |
107 | for (Element commentElement : listing.children()) { | |
108 | if (commentElement.hasClass("comment")) { | |
109 | Comment comment = getComment(commentElement); | |
110 | if (!comment.isEmpty()) { | |
111 | comments.add(comment); | |
112 | } | |
113 | } | |
114 | } | |
115 | return comments; | |
116 | } | |
117 | ||
118 | private Comment getComment(Element commentElement) { | |
119 | String title = firstOrEmptyTag(commentElement, "h3"); | |
120 | String author = firstOrEmpty(commentElement, "h4"); | |
121 | String content = firstOrEmpty(commentElement, "comment-body"); | |
122 | ||
123 | String date = ""; | |
124 | int pos = author.lastIndexOf(" on "); | |
125 | if (pos >= 0) { | |
126 | date = author.substring(pos + " on ".length()).trim(); | |
127 | author = author.substring(0, pos).trim(); | |
128 | } | |
129 | ||
130 | Comment comment = new Comment(commentElement.id(), author, title, date, | |
131 | content); | |
132 | ||
133 | Elements commentOutline = commentElement | |
134 | .getElementsByClass("comment-outline"); | |
135 | if (commentOutline.size() > 0) { | |
136 | comment.addAll(getComments(commentOutline.get(0))); | |
137 | } | |
138 | ||
139 | return comment; | |
140 | } | |
141 | ||
142 | /** | |
143 | * Get the first element of the given class, or an empty {@link String} if | |
144 | * none found. | |
145 | * | |
146 | * @param element | |
147 | * the element to look in | |
148 | * @param className | |
149 | * the class to look for | |
150 | * | |
151 | * @return the value or an empty {@link String} | |
152 | */ | |
153 | private String firstOrEmpty(Element element, String className) { | |
154 | Elements subElements = element.getElementsByClass(className); | |
155 | if (subElements.size() > 0) { | |
156 | return subElements.get(0).text(); | |
157 | } | |
158 | ||
159 | return ""; | |
160 | } | |
161 | ||
162 | /** | |
163 | * Get the first element of the given tag, or an empty {@link String} if | |
164 | * none found. | |
165 | * | |
166 | * @param element | |
167 | * the element to look in | |
168 | * @param tagName | |
169 | * the tag to look for | |
170 | * | |
171 | * @return the value or an empty {@link String} | |
172 | */ | |
173 | private String firstOrEmptyTag(Element element, String tagName) { | |
174 | Elements subElements = element.getElementsByTag(tagName); | |
175 | if (subElements.size() > 0) { | |
176 | return subElements.get(0).text(); | |
177 | } | |
178 | ||
179 | return ""; | |
180 | } | |
181 | } |